How Gujarat Results Shredded The Credibility Of Many A Pollster

From six months before the assembly elections in Gujarat, Swarajya had repeatedly cautioned polling agencies and media houses on two points:

One, their ‘tendency’ to inflate projected vote shares for the Aam Aadmi Party (AAP); and, two, their ‘finding’ that such an AAP wave could come to pass without the party having set up any robust state-wide grassroots network, and in spite of no ‘wave’-level electioneering visible on the ground.

If the pollsters were right, then the AAP was set to establish a new paradigm in Indian politics — of a party’s ability to sweep an election solely on the basis of a media blitz, and without an elaborate organisational structure.

Swarajya’s friendly advice was based on past precedents from Goa and Uttarakhand, where too, in the assembly elections of March 2022, they had over-estimated the AAP vote share by orders of magnitude.

In a business where the acceptable standard margin of error cannot be more than 3-5 per cent, these agencies blithely went off on reckless tangents.

In Uttarakhand, they said that the AAP would poll between 11-15 per cent; the reality is that the AAP got only 3 per cent, and no seats.

It was the same in Goa, (worse, in fact) where multiple polls said that the AAP would get a quarter of the popular vote and win at least six assembly seats.

But when the results arrived, we saw that the AAP got only 6.8 per cent of the vote and just two seats (one of which was, in reality, a candidate win rather than a party one).

Thirteen per cent (the midpoint of the Uttarakhand opinion polls) versus 3 per cent means that these agencies were wrong by a whopping 77 per cent, and off by 10 percentage points.

Similarly in Goa, predictions of 23 per cent for the AAP (the midpoint of the surveys), against a reality of 6.8 per cent, means that these agencies were wrong by 71 per cent, and off by a staggering 16 percentage points.

But no one spoke, and no one cared, because the mission had been accomplished — the creation of a media buzz around a non-existent AAP wave.

And yet, in spite of all that, these pollsters and media houses persisted in their claims that the AAP would garner over 20 per cent of the vote share in Gujarat, and win about one or two dozen assembly seats.

In contrast, both Swarajya and Zee24 chose to sidestep the hype and buzz to make their own, quite different, forecasts. A comparative table is presented below:

Even as leading pollsters and media houses continued to maintain that the Bharatiya Janata Party’s (BJP) vote share would decline from the 49 per cent it received in 2017, our own internal assessment was that the BJP had already crossed the 50 per cent mark in 2021, through a series of by-elections.

Consequently, our base forecast (and hence conservative by nature), was that the BJP would win at least 125 seats with a minimum vote share of 50 per cent.

While it was clear that the BJP would do much better than this base forecast in reality, precise quantification was not possible in the absence of a survey.

Therefore, a qualitative forecast was presented, in which, there was a high probability of the BJP crossing 135 seats, and of even registering a Rajasthan 2013-like sweep.

This assessment did not expect a decline in the BJP’s vote share, or any major state-wide shift of BJP votes to the AAP.

This went against those pollsters who said that the BJP, the ‘Others’, and the Congress, would all lose votes to the AAP, and only to the AAP.

Regarding the Congress, it was clear that they would lose a chunk of their votes, mainly to the AAP and some to the BJP.

The extent of that vote erosion was unclear, but it was qualitatively factored in to our assessment thus: the more the AAP cut into the Congress’s votes, the better the BJP would do.

Hence the likelihood of a sweep à la Rajasthan 2013, when the BJP won 163 seats out of 200 (82 per cent).

In addition, and representative of ground realities, word on the street was that Congress firebrand Jignesh Mewani would be hard-pressed to hold on to his Vadgam seat in North Gujarat (he won in the end, but only after trailing for many rounds, and by a margin of merely 4,928 votes).

Considering the above, the chances of the AAP making it into double digits, on both vote shares and seats together, appeared bleak.

At best, they might do well in pockets — like in a few tribal seats, or in those where the Muslim vote shifted to the AAP, or in some rural seats of Saurashtra, where legacy disaffection against the BJP persisted from Keshubhai Patel’s short-lived rebellion of 2012, and the Congress’s three-pronged caste ploy of 2017.

But even those added up to less than half a dozen seats. And pertinently, that list did not include the AAP’s more prominent Gujarati faces like Isudan Gadhvi or Gopal Italia who were expected to lose (they did).

This assessment was bolstered by the absence of any serious campaigning by the AAP in the cities; and this was after the party’s leader, Arvind Kejriwal, had scowled his way past the Ahmedabad Police for a widely-publicised autorickshaw ride.

The proof, albeit ex post facto, was that AAP did not even have a voting stall at this writer’s designated polling station in western Ahmedabad (the BJP polled 83 per cent in this seat).

And in the end, the results showed that the polling agencies were so off the mark, on so many counts, that only politeness prevents us from commenting more on them.

The BJP increased its vote share to 53 per cent, won 156 seats out of 182 (86 per cent) to best their sweep of Rajasthan in 2013, and set a record for the maximum number of seats won by any party in any assembly election in the history of Gujarat.

The Congress may take Pyrrhic solace in the fact that their record of maximum vote share remains unbeaten — 55.6 per cent under Madhav Singh Solanki in 1985.

The AAP won just five seats — four in Saurashtra, and one on the tribal belt. Of these, they gained two from the Congress, one from Chotu Vasava’s Bharatiya Tribal Party, and just two from the BJP. Even their final vote share of 13 per cent is actually an inflated figure because voter turnout dropped drastically by 7-10 per cent from 2017.

Now that the Gujarat and Himachal Pradesh elections are over, and all eyes turn to Karnataka, we may only hope that polling agencies and media houses take the opportunity to fully redeem themselves.

India has changed, and the days of taking poll predictions by famous names at face value are long gone.

Today, Panglossian pollsters who try to push implausible narratives through questionable surveys will be called out on the spot, and their credibility will be irreversibly affected.

All electoral data from Election Commission of India website.